snowball-67b-a2b-sft-s3-opencode-step1888

Snowball 67B-A2B SFT Stage 3: OpenCode

This is the BF16 Hugging Face/vLLM export of the OpenCode stage in Marin's ordered Chat → Thinking → OpenCode SFT campaign. It was trained on OpenCode trajectories and should be evaluated with an OpenCode harness.

main is the final step-1,888 checkpoint. The retained step-1,000 checkpoint is published on the step-1000 branch.

Training provenance

• Initialization: final Stage 2 Thinking checkpoint at step 630
• Data: tokenized/[email protected], using the restored fixed-EOT path from marin-community/marin#8171
• Data lineage: 29 OpenCode trace datasets generated with a Qwen3.5-122B teacher
• Schedule: five token epochs, 1,888 steps, sequence length 32,768, global batch size 64
• Final loss: approximately 0.201
• Native checkpoint: s3://marin-us-east-02a/marin/grug/snowball_step105149_sft_s3_agentic_eot_5ep/2026.08.13.1/checkpoints/step-1888/
• Recipe source: marin-community/marin#8172 at 7b1ea1dc07
• Runtime source: the recipe above plus distributed checkpoint writing from marin-community/marin#8140 and a two-hour checkpoint barrier configured for this run
• Experiment: marin-community/marin#8225

The export preserves the Marin tokenizer and Delphi special tokens. It is not the Nemotron-Terminal/Terminus-2 Stage 3 described in the successful control from marin-community/marin#7743.